New Style Rules discussion thread

[Leeghoofd]

I would sincerely ask you to tone down Yosi, show some respect. 

The latest Cat version (ver 2.0) runs pretty smooth in my book.  And about the glitching, that  in my book is usually user imposed, the same that happened with many other benchmarks. We rate them unsafe, glitchable,... the problem is many times in the chair, not in the computer...

If you want to debate, do it at an adult manner and drop the cocky attitude...

[Mr. Fox]

On 2/2/2021 at 1:25 PM, rbuass said:

I would like to do a resquest, to your analisys.

There are 2 special benchmarks that I think should be included for Global.

 

One is GPUpi 1B, because will show the highest frequency some graphic card can run a test.

GPUpi 32B is heavier, and so, the overclocker cannot run with too high clocks, so is very impressive to see running @ 3 Ghz or more.

 

The second is Catzilla, that's a very nice 3D bench allowed for Global points in the past (2 versions, 720P and 1440P).

It would be great to see Catzilla 1440P again on ranking for Global points, since requires very strong performance and tweaks.

Just for reference, Allion Labs, I think one of the most important certification labs in the world, uses Catzilla for tests, since will have very strong requests and peaks.

Who remember the crashs under Raymark, know what I'm talking about.

 

Best wishes for all overclocker.

 

 

On 2/3/2021 at 4:33 AM, yosarianilives said:

Lmao calling catzilla a nice bench, it sounds like you've never tried to run it. It's a glitch fest and we all know it, not to mention a bitch to run. 

Actually, the latest Catzilla update is not very glitchy at all. It runs great. On top of that it is entertaining, so there is a plus. It actually pulls more watts from my 3090 Kingpin GPU than any other 3D benchmark.  It used to be as buggy as hell, so I know why you are saying that. Truth be told, the garbage UL/Futuremark is putting out is about a buggy as anything. There is constantly something goofed up with their junk DLC in the 3DMark suite. VR Mark has never been updated and most of the time it completes successfully with no score, exits to the desktop with no useful information whatsoever about the error, or something else. 

https://youtu.be/69Lx6zLztHs

Edited February 5, 2021 by Mr. Fox

[rbuass]

On 2/2/2021 at 6:02 PM, Leeghoofd said:

Just propose it in the 3D thread Ronaldo. This thread is about the rules...

 

Users found GPU1B too fast, (yes you can claim +3GHz core clocks)

Users nearly crucified me for wanting to introduce Catzilla 4K, but I'll leave it up to you to persuade them to reconsider  

No, mate, I consider that you do a great job in the league, Indispensable for the organization of the HWBOT.
Sorry if I posted in the wrong place, but I really find these two benchmarks interesting for the reasons I mentioned.

In any case, we are a community and I understand that the majority should prevail.,

[Mrmouse]

Hi Guys ?

Glad to see some Upgrade on HwBot; the Rules look more clean now and should be more overseeable to new Members.
Fine Stuff and i hope for some positive HwBot-Processing this Year ?

What me wonders a bit; why is 'CPU-Z Mainboard' no longer mandatory?
In some Cases this could be very interesting and, in my humble Opinion, it is an important Aspect in Overclocking.
(To open CPU-Z a third time is really no Effort)

And what is with GPU-Z Monitoring?
Is it now Mandatory? Or is the first GPU-Z Tab enough?
Older Cards with AGP and Stuff like that; do they need Monitoring?

Best Greetings,
mrmouse

[Leeghoofd]

CPUZ monitoring refers for cards with boost clock, so we have a better idea for the real clocks. Usually doing a render via GPUZ after the run is fine. But its not mandatory

CPUZ for mainboard is only used in competitions sir.

[max1024]

I have a couple of thoughts. We are talking about motherboards with several physical sockets, 2, 4, 6 and 8. How many windows should be opened in the CPU-Z (options - one with the last processor's number or Two - with the first and last processor)? I think something needs to be written in the rules for multi-socket configurations.
The second thought concerns the accrual of points & etc. for the CPU-Z validation of such multi-socket systems. The current version of the Rules does not provide any score for this, since the best result is taken from one-socket configurations. But why not mention the efforts of the owners of EVGA SR-2 and other motherboards?, who overclocked their Xeons in pairs? Yes, overclocking will be weaker than on single-socket systems, but the excitement is higher, as well as the complexity of such overclocking. Think about this proposal.

And a little more in defense of the CPU-Z tab for the motherboard, in my opinion this is important and even useful, since sometimes it is not clear on which board the result was made, and if it was with which version of the BIOS, this is useful when you have overclocked problems , and a simple solution is to look at the hwbot for similar results of this motherboard. Imho double benefit in one word.

[Mr.Scott]

FWIW, I've been doing CPU tabs for both sockets on all my SR-2 subs since the begining. Figured it was just a common sense requirement.

[Leeghoofd]

Added the most important changelogs to the new rules... if any oversight let me know plz

[Rauf]

17 minutes ago, Leeghoofd said:

Added the most important changelogs to the new rules... if any oversight let me know plz

Been wondering for some time about some 3dmarks. In the past it was stated that sysinfo could be disabled for 3d03, 05 and 06. It seems it's not allowed anymore for 05 and 06? 3d03 does not require validation link anymore so I guess it's ok to disable sysinfo. 

For cloud gate and sky diver the vast majority of scores are done with tess/lod. But it's not allowed anymore.

Both of these changes messes with the hw points as it has a performance hit. It's not really a level playing field anymore.

Also wondering about the necessity of valid links for 05 and 06, if we use tess/lod the links are not public anyway. Is it really meaningful to provide another screenshot of the validation page? It's just two screenshots instead of one, it does not add a layer of security.

[Leeghoofd]

All these benchmarks scale tremendously with the subplatform therefore we opted to go for Sky to follow UL. Same rule set for almost the entire 3DMark suite, so VALID verification links required, meaning no LOD nor TESS. this might not even the playing field versus older scores but will do so for future scores/hardware

For 05/06 its only for the global top 20 that we need a verification link, not a VALID link. Systeminfo stays enabled. If needed the mods will inquire after the screenshots of the validation page if hidden....

[Antinomy]

@Leeghoofd, what's the position on old results being submitted? I've found a small pile of socket 7 results that haven't been published and are about 8 years old. CPU-Z is outdated on those obviously. Not that it changes anything on 25 year old hardware.

[Leeghoofd]

Use "the backup on USB drive comment" Alex

[unityofsaints]

On 4/24/2021 at 2:59 PM, Leeghoofd said:

All these benchmarks scale tremendously with the subplatform therefore we opted to go for Sky to follow UL. Same rule set for almost the entire 3DMark suite, so VALID verification links required, meaning no LOD nor TESS. this might not even the playing field versus older scores but will do so for future scores/hardware

For 05/06 its only for the global top 20 that we need a verification link, not a VALID link. Systeminfo stays enabled. If needed the mods will inquire after the screenshots of the validation page if hidden....

Good to know it doesn't need to be valid, heard tons of complaints from other benchers saying it's hard to get a valid out of Ryzen on coldwith Windows 7, regardless of HPET on or off (haven't tried myself yet).

[Leeghoofd]

Link verification rules have been dropped for 05/06 since end of May, always check the latest rules J

[unityofsaints]

AS SSD gets globals now but doesn't have a rules page yet, XTU 2.0 gets hwpts but also doesn't have a rules page. Are these in the works?

[Leeghoofd]

ASSD had a rules page (updated the screenshot)

added XTU 2.0 -watchout for the globals screenshot requirement.

[Leeghoofd]

Most benches from the 3DMark suite have standard rules in compliance with UL, no more changes required